The present invention relates to an image processing apparatus and method and, more particularly, to an image processing apparatus and method for performing color processing of an input image signal.
An example of a conventional color image processing apparatus will be described below with reference to FIG. 12 taking color image processing in a color copying machine as an example. In FIG. 12, reference numeral 101 denotes a color image input unit such as an image reader unit of a color copying machine. The color image input unit outputs three color-separated signals R1, G1, and B1 which are obtained by color-separating each pixel of a color image into R, G, and B components. The three color-separated signals R1, G1, and B1 are input to an achromatic color/chromatic color determination unit 1201, which determines if the pixel of interest is a monochrome pixel (achromatic color) or a color pixel (chromatic color), and outputs a determination signal KC.
The signal G1 of the three color-separated signals is input to a character/halftone image determination unit 111, which determines if the pixel of interest corresponds to a line image such as a character or a thin line or a continuous-gradation image (halftone image) such as a picture image or a printed image, and outputs a character/halftone image determination signal TI. The character/halftone image determination signal TI is input to a spatial filter coefficient storage unit 112, which selects character spatial filter coefficients 1301 shown in FIG. 13 when the pixel of interest corresponds to a character signal, or selects halftone image spatial filter coefficients 1302 shown in FIG. 13 when the pixel of interest corresponds to a halftone image signal.
Conventional spatial filter processing such as edge emphasis and the like will be explained below. FIG. 13 shows an example of the 5×5 pixel character spatial filter coefficients 1301, and the halftone image spatial filter coefficients 1302. The character spatial filter coefficients 1301 are determined to effect stronger edge emphasis than the coefficients 1302 for a halftone image.
The character or halftone image spatial filter coefficients Kij selected in accordance with the character/halftone image determination signal TI are set in edge emphasis units 103-R, 103-G, and 103-B to respectively edge-emphasize the three color-separated signals R1, G1, and B1, thus outputting signals R2, G2, and B2.
FIG. 14 shows an example of the edge emphasis unit 103-R. A dotted frame 1401 represents a data delay circuit in the edge emphasis unit 103-R. The signal R input to the edge emphasis unit 103-R is input to line memories 861, 802, 803, and 804 that store image data for four lines. Image data for a total of five lines, i.e., the stored image data for four lines and image data of the line of interest, are input to flip-flops in units of lines to extract five successive pixel data (Xj1 to Xj5).
The signals R for 5 lines×5 pixels, i.e., a total of 25 signals R (Xij: 1≦i≦5) are multiplied by spatial filter coefficients (aij: 1≦i≦5, 1≦j≦5) indicated by a dotted frame 1403 and corresponding to pixel positions by an edge emphasis arithmetic circuit indicated by a dotted frame 1402, and the products are added to each other. These spatial filter coefficients are supplied from the spatial filter storage unit.
Such calculations require 25 multipliers (1404 to 1428) for 25 pixels, and 24 adders (1429 to 1452) of the products.
The conventional spatial filter processing for the signal R has been described. The same applies to signals G and B. Consequently, the processing circuit for all the signals R, G, and B requires a circuit scale: line memories for 12 lines (4 lines×3 colors), 75 multipliers (25×3 colors), and 72 adders (24×3 colors).
The three edge-emphasized color-separated signals R2, G2, and B2 shown in FIG. 12 are input to a luminance/density conversion unit 106, and are converted into density signals C1, M1, and Y1 by, e.g., log conversion. The density signals C1, M1, and Y1 are input to a color correction unit 107 to be subjected to color processing such as generation of a black signal K, undercolor removal (UCR), color correction, and the like, thus outputting density signals C2, M2, Y2, and K2.
The color correction unit 107 sets the density signals C2, M2, and Y2 at C2=M2=Y2=0 in accordance with the determination signal KC as the determination result of the achromatic color/chromatic color determination unit 1201 when the pixel of interest is an achromatic pixel, thereby converting the pixel of interest into a pixel defined by black color alone. Reference numeral 110 denotes a color image output unit which comprises an image recording apparatus such as an electrophotographic or ink-jet printer.
When the color image output unit is, e.g., a binary printer, the density signals C2, M2, Y2, and K2 are converted into binary pixel signals C3, M3, Y3, and K3 by a binarization unit 108.
On the other hand, when the resolution of the image input from the color image input unit 101 is different from that of the image to be output from the color image output unit 110, the binary pixel signals C3, M3, Y3, and K3 are subjected to resolution conversion processing by a smoothing/resolution conversion unit 109 to be converted into signals C4, M4, Y4, and K4. Especially, when the resolution of the color image output unit 110 is higher than that of the color image input unit 101, smoothing processing for smoothly interpolating edge portions of the image is performed, and the processing result is printed by the color image output unit 110.
However, as described above, the conventional image edge emphasis requires identical arrangements in units of signals to process the signals R (103-R), G (103-G), and B (103-B). Especially, in order to perform two-dimensional plane spatial filtering processing of the above-mentioned 5×5 pixel size for an image signal from the image reader unit of the color copying machine using a CCD line image sensor as the image input unit, many line memories, multipliers, and adders are required, as described above, resulting in high cost.
In terms of image quality, since the above-mentioned color correction unit 107 sets the density signals C2, M2, and Y2 to be C2=M2=Y2=0 in accordance with the determination result of the achromatic color/chromatic color determination unit 1201 when the pixel of interest is an achromatic pixel, so as to convert the pixel of interest into a pixel defined by black color alone, the densities obtained by the signals C2, M2, and Y2 are lost, resulting in a low density.
In general, in a color image processing apparatus, especially, in a color copying machine, when an image corresponding to a monochrome original is to be formed and output, the image is copied using four colors, i.e., C (cyan), M (magenta), Y (yellow), and K (black). However, in the case of a laser beam printer, in consideration of the service life of a drum and consumption of toner, the image corresponding to the monochrome original is preferably copied using black color alone. The same applies to a copying machine that incorporates an ink-jet printer.
For this reason, the copying machine is required to have a processing unit for determining if the input original is a color or monochrome original. Conventionally, such processing is realized by simple evaluation, i.e., by summing up the color pixels of an input original and performing simple statistical processing of the sum or comparing the sum with a threshold value or a slice level.
However, when color pixel determination is done based on color components (in this case, R, G, and B luminance values) in units of pixels in an original input device, if the reading elements for the individual color components cannot read perfectly the same pixel positions in an original, i.e., they are shifted by a small distance, a so-called pixel shift phenomenon occurs, and this finally results in. “color misregistration”. In particular, since recent image scanners tend to have higher reading precision (resolution), the above-mentioned problem is inevitable.
As can be understood from the above reason, when an image scanner reads a uniform original (e.g., white paper), color misregistration does not show even if the reading position of the image scanner has shifted.
The problem is noted in the vicinity of the edge in, e.g., a black character or line image (to be referred to as a black character hereinafter). If color misregistration has occurred at the edge of a black character due to pixel shift, the pixel at that position is determined to be a chromatic pixel. Hence, the above-mentioned simple statistical processing or comparison between the number of pixels and the slice level may determine a color original if the original includes many characters.
As a consequence, an image that must be copied using black color alone is copied using C, M, Y, and K.
On the other hand, a specific color is likely to be determined as an achromatic color, especially, in the case it has a high density.
Furthermore, even when an achromatic color is read by a scanner, color misregistration is caused by mechanical vibrations, chromatic aberrations of lenses, different MTF, and the like, and the read achromatic color may be erroneously determined as a chromatic color.
Since these determination errors associated with achromatic/chromatic colors have adverse influences on the subsequent processing in the color correction unit and edge emphasis units, the quality of the image to be finally formed deteriorates.